Pricing Interest Rate-Sensitive Credit Portfolio Derivatives
Weird Science: Dissection of Derivatives · 2011. 11. 3. · l'm going to go over the current usage...
Transcript of Weird Science: Dissection of Derivatives · 2011. 11. 3. · l'm going to go over the current usage...
RECORD, Volume 24, No. 1*
Maui I Spring Meeting
June 15-17, 1998
Session 45OF
Weird Science: Dissection of Derivatives
Track: lnvestment
Key Words: lnvestments, lnsurance Companies
Moderator: STEPHEN D. REDDY
Panelists: CHRlSTOPHER T. ANDERSONt
CHARLENE MARlE BARNES
LARRY M. GORSKl
Recorder: STEPHEN D. REDDY
Despite negative press and numerous instances of imprudent use of derivatives and
the structured products that embody them, the demand for derivatives continues to
increase and the suppliers are more than willing to accommodate. However, an
ever-growing problem and challenge is how to regulate and account for all these
instruments properly in the context of their multiple applications.
Mr. Stephen D. Reddy: We have an excellent panel for this session. Charlene
Barnes is an ASA and assistant vice-president with General Re Financial Products.
Charlene specializes in derivatives and structured products, with an emphasis on
insurance asset liability management and insurance tax issues. Larry Gorski has
been a life actuary with the lllinois Department of lnsurance for 23 years and chairs
the NAlC's lnvested Asset Working Group and Risk-Based Capital Working Group.
He's a member of the NAlC's Life and Health Actuarial Task Force and has been
active in several Academy task forces, including the Equity lndex Product Task
Force and the Valuation Task Force. Chris Anderson is Merrill Lynch's specialist in
investments by insurance companies. He works closely with senior investment
officers and chief investment officers (ClOs) and has done a substantial amount of
work supporting the industry on regulatory matters.
*Copyright © 1999, Society of Actuaries
†Mr. Anderson, not a member of the sponsoring organizations, is Director, Fixed Income Research of Merrill
Lynch & Company in New York, NY.
Note: The charts referred to in the text can be found at the end of the manuscript.
2 RECORD, Volume 24
Ms. Charlene M. Barnes: l was wondering if this should be called "Mad Science
and the Dissection of Derivatives," given the kind of general view that a lot of
people have about derivatives.
The first question people want to know is, "Why use derivatives?" One big reason
not to use them is what we call "Wall Street Journal risk." These are very powerful
tools, and nobody wants to be the person responsible when they blow up. But
there are also a lot of very good reasons to use derivatives. Cutting yourself off
because of lack of understanding can be very costly for your company.
Derivatives can help you manage investment risk by decreasing the volatility of cash
flow or earnings and reducing the risks of surplus drain. You can also generate tax
savings from derivatives. For example, if you have positions in place and want to
change their duration but have a lot of embedded capital gains in your existing
portfolio of bonds, you can use derivatives to change duration without having to sell
the bonds or sell them and generate a capital gain. As long as you are within
certain risk parameters of the lRS, that's not a problem.
Derivatives are also used extensively for asset and liability management. You can
instantaneously and cheaply shorten or lengthen the assets to match the liabilities
better. You also can change the convexity of your asset portfolio. Most of the
things we consider natural assets are more sensitive to interest rate changes than are
liabilities. Derivatives can help even that out. They also allow you to do asset and
liability management without actually disturbing the assets that you own, which you
want to avoid for tax reasons. Also, if you have, for example, a money manager
who's doing very well in short-term money management but you want a longer
duration, you can use derivatives to lengthen the duration of your portfolio without
punishing that money manager by taking money away from him or her.
Another big reason to use derivatives is that, like it or not, you're already using
them. Most insurance products already have derivatives embedded in them. For
example, with minimum nonforfeiture benefits, you're essentially writing a put.
With minimum interest rate guarantees, you're writing a floor; no matter what
happens to the interest rates, you're going to credit 3%. Equity indexed products
are another very obvious example of derivatives that are embedded in insurance
products.
The question shouldn't be, "Why use derivatives?" but "How are you going to
hedge the derivatives you're short?" lf you're not purchasing derivatives, you have
to have some plan in place to adjust your asset mix if interest rates change
dramatically. Basically, you want to use your entire portfolio and create a dynamic
hedging scheme to hedge the derivatives that you're short.
3 Weird Science: Dissection of Derivatives
l'm going to go over the current usage of derivatives, actuarial pricing versus
arbitrage free pricing, and emerging markets in derivative products. l'm not going to
talk about structured products, although it's a very interesting topic. l'm not going
to talk about accounting, which is a hot potato right now. l'm not going to talk
about taxes, because taxes of derivatives fall into two camps: the extremely simple
and extremely complex. And l'm also not going to talk about pricing models such
as Black-Scholes and Monte Carlo. Suffice it to say that, in general, with derivatives,
there's either a nice, neat formula that will act as a model, or you can use a Monte
Carlo simulation. However, you can't use a Monte Carlo simulation if it's not
hedgeable, and l'll get into that in a bit.
For current usage estimates, l'm using the 1996 Schedule DB, which is where all
true derivatives live in your statements. lt's not a great estimate because it's based
on notional amounts. The notional amount is an indicator of how much of a
derivative you have, but it's not a very good indicator of the actual risk or cost
involved in these derivatives. However, we've estimated that about 40% of the
usage, or $49 billion notional, is in swaps with insurance companies, and $74
billion, or 60% of the total notional, is in other types of options owned by insurance
companies. Chart 1 shows the largest insurance company users of derivatives. lt's
somewhat complicated, but gives you a lot of information about who's actually
using derivatives. The bars show the total notional for the top 10 life insurance
companies using derivatives. The dotted line uses the axis on the right to show the
percent of the total that each of those companies represent. lt is directly
proportionate to the bar. The solid line shows the cumulative percentage of the
total. The first company uses about 8% of the entire notional derivative. Then it
creeps up quickly. The top 10 users of derivatives account for 70% of the
derivatives that are being sold.
Chart 2 shows off-balance sheet applications. Note that annuity hedging accounts
for 53% of the applications. This covers disintermediation risk, which is often
covered in swaps, guaranteed minimum death benefit, the S&P 500 exposure and
equity indexed annuities, and negatively convex assets. The shape of the asset and
the shape of the liability over different interest rates is very different, and that
represents people using derivatives to take care of that problem.
Twenty-four percent use derivatives for GlC hedging, which is primarily duration
risk, embedded optionality, and basis risk. The final 23% fall into other categories:
structured settlements, which are very elongated liabilities; life insurance
applications that are similar to annuity risk, but not quite as emphasized in the
product and the profitability of the product; and asset replication-equity-linked
notes or creation of different types of exposure using derivatives.
4 RECORD, Volume 24
Actuarial pricing is a historical experience. We have an enormous amount of
experience for actuarial pricing relative to the amount of experience we have in the
capital market, because no matter how many pieces you divide capital market
history into, you still have only one capital market. Actuaries are used to working
with millions of independent lives.
With actuarial pricing, changes creep in. Mortality rates improve-although they
did get worse during the AlDS crisis-but this doesn't happen instantaneously. lt
moves into the population or moves out of the population over a period of at least a
couple of years. And, the lives are all independent. One can argue that they're not
completely independent, but it's a very fine point.
Actuarial pricing involves a lot of judgment. ln looking at the historical data, you
have to decide what's happening. Mortality improvement factors are just
quantified judgments. Finally, with actuarial pricing, when you're coming up with a
price or product, you need a satisfactory risk/return relationship for the company.
Often that's expressed in a very simple return on equity type of analysis. But you
recognize that you have some risk and need to get a better return than you'd get by
investing in treasuries.
Actuarial pricing differs from arbitrage free pricing, which is used to price most
derivatives. Arbitrage free pricing is determined by totaling the current market price
of all the components. l like to call it "selling the product for parts." You determine
the risks, the cost of those risks in the current market, and then add them all
together. Experience is irrelevant because you're basically going to take on some
risk and sell off the risk components. The risk/return component is irrelevant
because, in true arbitrage pricing, you're not keeping any risk, so you're not going
to get any return for it. Lower risk, in general, means lower return for these
products. So arbitrage free price product involves very simple derivative, like
swaps.
A simple insurance example illustrates the difference between actuarial pricing and
arbitrage free pricing. On the actuarial side, you might want to set the premium (P)
today for a risky payment (X) in five years. X is a random variable; you don't know
what it is, but it will be something in five years. The formula is P-Xv5, or P minus X
times the discount factor, which you want to equal zero or some acceptable
profitability level. Using historical analysis, current trends, conservatism, and profit
you come up with the acceptable expectations for X. Ultimately, P is a function of
the risk tolerance and risk/return analysis. You're taking on some risk and getting
paid for it, so that's how you determine the P portion of the formula.
5 Weird Science: Dissection of Derivatives
Let's take basically the same example with a forward sale paid in advance. You
want to find the price today (P) to sell a single share of stock in five years. Once
again, there's a certain return on the stock, so it will be worth something else in five
years. You could try to determine that number using historical analysis, reasonable
expectations, and judgment, but, in this case, those don't make the slightest bit of
difference. lnstead, you collect Y dollars today, buy one share, and then give it to
the counterparty in five years. That's all there is to it. The price is equal to the price
of the share plus some profit (P = Y + profit). Because this is essentially a risk-free
transaction, that profit will be a relatively small amount.
Arbitrage will always drive prices to the arbitrage free pricing level. Let's say that,
instead of the simple insurance example, you were insuring the price of a share in
five years. You have two different pricing methods. Because somebody can
arbitrage this and take no risk for a very slim profit, the prices are going to have to
converge at some point. And the only real difference between the two prices is a
risk/return analysis. lf the actuarial price is too high, the market won't buy it. You'll
have to reduce the return for the amount of risk you're keeping. lf it's too low, you
could sell this risk in the market and get a lot more money, so what are you doing
selling it in an insurance product?
We're working with the efficient market theory, which says that the market's
risk/return is the best estimate. Whatever the insurance company thinks the
risk/return relationship is doesn't matter. The price of an asset such as a share of
stock has the risk/return of the entire market embedded into it. And that's the
risk/return relationship that matters. lt's not an unbiased estimate, but it's the best.
Finally, certain risks cannot be arbitraged because there is no market. lf you can
buy that single share of stock in the market, you can get rid of all the risk and use
arbitrage free pricing. lf you could not buy that share in the market, for example, if
there was some other sort of risk, like a life insurance risk, then arbitrage free pricing
just doesn't work.
There is a market for your life insurance risk in reinsurance. And that's almost pure
arbitrage, but not quite, because it's a one-way market for the most part. You can
sell your life insurance risk to a reinsurer, but usually can't buy it. Within the
reinsurance and retrocessionnaire market, there is some transfer back and forth. But
for a direct writer's purposes, it's a one-way market.
There's some risk sharing with reinsurance. Very often the insurance company
doesn't want to lay off the entire risk to the reinsurer. lf, for example, the company
is developing a new product, it would want some risk sharing to ensure that
everyone is doing due diligence and giving their best effort.
6 RECORD, Volume 24
But if you sold all your risk to the reinsurer, it would be a spread business. You'd
bring it in for a certain price, arbitrage it, sell it for parts for a lower price, and keep
a bit of money. That's basically how arbitrage works. The problem is, you reduce
the risk and the return, so is it worth it? You're selling the risk for a lower price than
you bought it for, but the reinsurers make the profit because they have a risk/return
relationship. The risk market being one-way poses the problem. ln a two-way
market, supply and demand would drive returns to the level that the whole market
finds acceptable.
Arbitrage free pricing principles allow you to value investments and investment
strategies. ln addition to valuing option transactions, the principles help you to gain
an understanding of market value and the sensitivity of the structured products.
Also, they are good for valuing unusual risks embedded in a security. As long as
you can arbitrage the risk on a share of stock or the bankruptcy risk on a company,
it can allow you to determine some value for it.
Also, arbitrage free pricing principles are very helpful in appropriate liability pricing
because they allow you to price embedded options consistent with market prices.
For example, with minimum interest rates in different types of annuities, you have
an embedded option in your product that you have to value. You can come up
with some sort of satisfactory risk/return relationship for your insurance company
based on history, but if the market is selling it for a whole lot more, you have to
look at all the different pieces of your product to make sure that you're not selling it
too cheap. Perhaps you could make more money by selling the risk directly, so it
doesn't eat up the entire profitability of the rest of the product. And even if you
don't explicitly hedge, keeping arbitrage free pricing methods in mind allows you to
get the risk/return relationship that's appropriate for the entire market.
Another issue is asset/liability management. Keep arbitrage free principles in mind
when you want to change your asset mix. lt will get you to a different point on the
risk/return relationship of the efficient frontier or move your entire company to a
different point. But you have to ask yourself, "ls this just pure arbitrage? Am l
moving to a better point or merely a different point?" lf you're moving to a point
that you could achieve simply by getting into a couple of option transactions, you
might not be making any big improvements.
Among emerging products, credit derivatives are very hot right now. Let me give
you a simple example. An insurance company wants to purchase a private
placement from corporation XYZ, but exposure to XYZ is already too high. And
let's just say that XYZ has publicly traded debt, which means there's a market for
their exposure and a potential to for arbitrage. With the credit derivatives, we have
a reference credit. XYZ has a public bond issue and floating rates with a maturity
7 Weird Science: Dissection of Derivatives
similar to the private placement. We'll use that as a reference credit when we're
creating a product. We'll also determine what credit event will trigger the payoff.
Usually it's a default of the reference credit.
So you select a bond and say, "lf that bond defaults, something is going to happen."
The insurance company then pays an annual premium to its counterparty, and, if
the credit event occurs, the derivative writer or the counterparty pays the difference
between the notional and the market value of the reference credit. Now you have
the private placement exposure to XYZ, and this derivative asset as well. lf XYZ's
publicly traded debt falls through the floor, you're still going to get some money.
Hedging your exposure allows you to buy more of the private placement. And the
level of protection depends on the terms of the private placement. Often private
placements will be more or less secure than the public debt. There's some basis
risk, but most of it is reduced by this type of transaction.
This example is hedgeable because of the public debt. The counterparty can short
the bond in the market, which means that, if the value of the bond goes down, the
counterparty will make money off the short and can pay that out. ln this example,
the market value of the reference bond is insensitive to interest rate changes, so that
won't be a factor. Also, the counterparty's risk and profit are relatively low. This is
very specific risk based on a reference credit that's actually available. The
counterparty is not taking any unusual risks, for example, the risk specifically related
to this private placement. Because it's arbitrage, the counterparty has laid off almost
all of its risk.
The more complex the product is, the less hedgeable it becomes. For example, the
public debt is of shorter term than the desired protection. You have a 20-year
private placement, and the longest term public debt is 10 years. That will make it
less hedgeable. Also, you can design it so the payoff is sensitive to interest rate
changes, which has nothing to do with credit risk. But it will still affect the value
because being less hedgeable means more risk for the counterparty. And more risk
means risk/return enters into the equation so it's more expensive from a purely
economic point of view.
To summarize, derivatives are an integral part of our business. Even if you don't
currently own any, they are embedded in your product, so you are hedging them in
one way or another. With arbitrage free pricing, the pricing for a hedgeable risk
differs from the pricing for a risk position. ln arbitrage free pricing, you sell off all
the risk and don't need any return. Actuarial pricing assumes you're keeping the
risk and need extra return for that risk. Most derivatives are priced without
risk/return analysis. When they become incredibly exotic, the counterparty does
maintain some risk, but also gets paid for that. lf there's no market for the risk,
8 RECORD, Volume 24
arbitrage free pricing does not apply. Finally, new products based on different risks
are constantly being developed.
Mr. Stephen D. Reddy: You showed that the top 10 companies are accounting for
about 70% of total derivatives usage in the industry. Why do you think it's as
heaped as it is? And what is the breadth of use beyond that 70%? Are a lot of
insurance companies not doing anything at all?
Ms. Barnes: There are a vast number.
Mr. Reddy: Why?
Ms. Barnes: lt's centered around a few companies right now because of the general
size of the company, the level of expertise, and when they made the commitments
to using derivatives. Obviously, derivatives come on and off the books, but some
are fairly long-term and there are 50-year swaps out there. The longer a company
has been using derivatives, the more they'll have. The other issue involves the
company's general level of expertise and the amount of risk they have. For
example, pure variable annuity writers don't have the same kind of risks as
companies that are writing fixed annuities. l find that more insurance companies
are developing expertise and interest in derivatives, though, so the proportion is
definitely increasing.
From the Floor: You list asset/liability modeling as one of your arbitrage free
pricing principles and ask, "ls another asset mix better?" Will you expand on that?
The way l've understood arbitrage free pricing is that, at any point in time, an asset
is fairly priced.
Ms. Barnes: That is correct. A simple example is if you have a particular asset
liability mix and you're planning to move to another asset liability mix. You might
think the risk/return of that new and improved mix is better than the current
risk/return combination. Arbitrage free pricing might indicate that it's not a better
mix. The difference in the risk/return relationship might have nothing to do with
asset/liability mix and merely be a point on the efficient frontier that our
stakeholders could have achieved themselves by shorting, fixing, buying equities, or
whatever. lf this is the case, you haven't accomplished anything so it's really not a
better mix. But if you move to a different point on the efficient frontier by changing
the asset mix alone, then you have definitely made an improvement.
Mr. Christopher T. Anderson: ln dealing with insurers, we've found a real
reluctance to get out front on this issue. Considering the efforts underway in the
9 Weird Science: Dissection of Derivatives
regulatory community in the last five or six years, it's been hard to find somebody
that's willing to take the lead and move forward.
What's so weird about this derivatives to begin with? What's so weird about
earnings pressures on companies? lf you look at the insurance industry in the U.S.
compared to western Europe, earnings are not as good. Performance pressure
continues in the investment department. ClOs care a lot about their performance
and what they can do about it.
What's so weird about seeking investment alternatives? That's been going on
forever, partially as a result of other factors. What's so weird about competitive
pressures for consumer resources? This is not the only industry that's vying for
funds. Rating agencies tell me that executive managers in this industry need to
understand that they are financial services providers on the life side and competing
for funds in the financial services area, too.
What's so weird about global competition? lt's not just your company in your city
anymore, and the pressures are significant. All of these pressures emphasize the
need for innovation. And when we talk about the weird science of derivatives,
that's what we mean.
Charlene went through some of the uses of derivatives. l'd like to make some
distinctions between structured products and derivatives. By a structured product, l
mean something that alters principal payment priorities. They see altered principal
payment streams, and periodic payment streams can and may be altered. The
classic example of a structured product and one of the early ones was the
collateralized mortgage obligation (CMO).
ln the insurance realm, a derivative product is one you first expect to see on
Schedule DB or in separate account. That's where to look. Derivatives include
swaps, options, futures, puts, calls, caps, and floors.
What happens in CMOs is that the principal repayment priorities are altered based
on the class of securities that an investor owns. They are allocated according to
prepayment characteristics spelled out before the transaction takes place. Also, it's
possible and likely that coupons will be reconfigured in a CMO. This is what we
mean by structuring in terms of altering the fixed and floating coupons and having
higher or lower coupons for various classes of security. You see structure not just in
CMOs but in collateralized bond obligations (CBOs), multiclass asset backed
securities (ABS), or basically anything that's multiclass and has been structured.
10 RECORD, Volume 24
What is important in your companies is not necessarily just risk-based capital (RBC)
numbers. My personal view is NAlC's risk-based capital formula is only of interest
to a very, very small number of companies. What is interesting, though, is the
classification of the security between debt, equity, and preferred. l currently have
an issue before the Security Valuation Office (SVO), which is making a
determination about whether it's a preferred or a common equity.
What makes a structured product bond-like? The SVO uses standards to determine
whether something is a bond or not, and the test is a predominance of these factors:
• Do creditor rights exist or not?
• Do you have a scheduled repayment of principal and interest? This is
mandatory. You have to pay or default. Lack of payment is an act of default.
• Are you a creditor in the event of liquidation or not?
• Are payments to holders deductible for tax purposes under U.S. law?
• ls the principal protected?
These standards were adopted a year-and-a-half ago, and l think it has helped all of
us distinguish debt from equity. lf a product meets all the conditions, the NAlC can
rate it as a bond, and under RBC of course, it's a bond. But, more important, the
rating agencies will view it as a bond.
lf the product is rated and monitored by a rating agency, we like that. We have
different standards for nonmonitored deals; for example, a security can be rated
once but currently not monitored. And you can have a product that was never rated
at all. We suggest getting products rated because it's much easier to get them to the
SVO.
A bond focuses on credit in the insurance realm. A rated credit instrument has a
specific meaning under the defined-limit model investment law. The law is
essentially the same as the one adopted in lllinois. lt means that there is no risk of
loss of principal due to factors other than credit. And that's a bond.
When you have a bond, we assume that no optionality exists, other than the
standard right of the borrower to prepay beginning at a certain time and at a certain
price. But there may be other optionality introduced by structure. l have to rely on
the actuarial profession to identify, evaluate, and measure those elements through
cash flow testing. The profession has done a terrific job of tackling CMOs. And the
regulatory community identified CMOs and gave regulators and others tools to deal
with them. We have to assume that the cash flow testing works, because with
11 Weird Science: Dissection of Derivatives
bonds, we're dealing with purely credit risk and other risk elements are not
identified.
Let's talk about a revised view of insurer use of derivatives. This is a legalistic
distinction, and l'm relying on the defined-limit version of the insurer investment
law. Uses for derivatives include: income generation and hedging. By income
generation, we mean writing covered calls. That's a very narrow, precise definition.
Hedging is a lot more complicated. ln this context, hedging means risk reduction.
You can hedge against a change in value, yield, price, quantity, or cash flow. You
can hedge changes on either side of the balance sheet, asset or liability, and they
can be either present or anticipated.
Now l'm going to forecast the past, because in GAAP accounting, the definition of
derivatives is something that fluctuates in price in accordance with other
instruments. The latest statement is just out, so l'm predicting the past. lf you want
a good prediction of the past, download an Ernst & Young (E&Y) paper called
"Derivatives and Hedging" from the Web site www.ey.com. Go to index,
accounting, and look under downloadable papers.
The concepts that E&Y expected to appear in the latest GAAP release include
discussion of derivatives as assets and liabilities and the directive that they should
be reported in the financial statements. But the key point is that fair value is the
most relevant measure for financial instruments and the only relevant measure for
derivatives.
Since codification of statutory accounting is unitary, it embeds what is rated as a
bond. Given the distinctions we've talked about earlier, a bond is rated based on
how it appears on the statement. Larry will be talking about that in terms of
replication and creation of synthetic assets, we see a major divergence for what we
expect between GAAP and STAT in this regard.
For GAAP accounting, only items that are assets and liabilities should be reported.
And finally, special accounting applies for items designated as being hedged. One
aspect of qualification should be an assessment of the effectiveness of the hedge
(i.e., offsetting changes in fair value or cash flows). This takes us to the next point,
which is how synthetic assets can be created using cash and derivative products.
Mr. Larry M. Gorski: My presentation today will be on the NAlC Replication
Project. Charlene explained that the current uses of derivative instruments were
annuity hedging, GlC hedging, and "other." Embedded in that other section was
something called "replication." Also, for the current uses of derivative instruments,
12 RECORD, Volume 24
she showed that the top 10 companies account for about 75% of total derivatives in
the insurance industry. l think it's an accurate assessment. From my standpoint, l
see about 65-70 companies using derivatives, but the vast majority of them are used
by nine or ten companies. And, if it wasn't for the introduction of equity indexed
annuities, there probably would be only about 45 companies using derivative
instruments. lt's still not a widely used instrument for asset liability management
purposes or risk management purposes in general. This is surprising, considering
the input we had from the industry and the trade associations when we were
developing the model investment law and the NAlC Replication Project. l would
have thought that there would be many more users, but there's still a great degree of
hesitancy in stepping forward as a significant user of the products.
lnsurers would like the flexibility to manufacture risk return profiles in a cost-
effective manner. That's the rationale for companies to use derivatives for
replication purposes. But regulators have some concerns. The first one is that the
current RBC formula assumes that derivatives are only being used for hedging
purposes. The only RBC factor associated with derivatives is for the counterparty
exposure. Because of that there are directives to move ahead with derivatives used
for replication purposes.
Second, the structure of reporting is very deficient when it comes to explaining the
use of derivative instruments. Schedule DB will identify derivatives by class
options, futures, forwards, etc., and there is some description there. But you get no
real sense of how the instruments are being used by the current reporting.
Third, regulators are concerned about any embedded bonds in a derivative
instrument. lt took six or seven years to develop the model investment law and,
during that period of time, there was one disaster after another. First we had junk
bonds. Then we had mortgage loans and the real estate crisis. And then we had all
the problems with stand-alone derivative instruments, such as the Orange County
situation and others. There was a great reluctance by regulators to move ahead with
recognition of derivatives by insurance companies. Chris did a good job of
explaining the three different uses from a regulatory standpoint: hedging, income
generation, and now replication.
What l'll be doing today is giving a brief overview of a report that has been
provided to the regulators by the interested persons, i.e., the industry. lf you're
interested in the whole report, l suggest calling Bill Shriner at the ACLl. The report
is still in draft form and changes will be made as time moves on. There's only one
element of the comprehensive project that is near completion, the reporting aspect
of it.
13 Weird Science: Dissection of Derivatives
A replication (or synthetic asset) transaction involves a cash market instrument and a
derivative instrument to replicate the performance of an otherwise permissible
investment. You might have a bond, a stock, futures, or a swap transaction. When
you use them as a single entity, they replicate or reproduce the performance of an
otherwise permissible investment. This is not to be confused with a structured note,
where both of the pieces are combined in a single instrument.
The key concept is reproducing the performance of an otherwise permissible
investment. That obviously implies some type of effectiveness measurement, but
developing that measurement has been the major stumbling block in the process of
recognizing derivatives used for replication transactions.
Lets consider a case where we're trying to replicate a lower-rated bond. We're
holding a U.S. government bond, and entering into a default swap on a specific
bond or issuer. From a cash flow standpoint, the bond is producing either fixed or
floating rate coupons. The insurer is receiving a semi-annual payment, but in the
event of default by the underlying bond or issuer, the insurer will be responsible for
the amount of that default. That's where the transformation of the government bond
into a lower-rated bond takes place.
Obviously there are variations to the structure. Rather than the swap being based
on a single bond or single issuer, it could be based on a basket or index and you
can have forward settlement. From the industry standpoint, the replication is
welcome because the item being replicated may not be available at the desired
maturity. lf you use a basket or index, you're bringing diversification as a
transaction. lt may be a more economical or more efficient transaction, so it will
have better value. And, because you're using government bonds as a component of
the transaction, they're obviously very liquid. The only illiquid piece is the swap
proportion of the transaction.
Here's another example. Let's say you want to replicate the performance of the S&P
500. You can buy treasuries and futures, combine them in the proper way, and
replicate the performance of a stock index. That's another example of a replication
transaction.
How are regulators going to account for these kinds of transaction? For the first
example-transforming a highly rated corporate or high-grade government bond
into a lower rated bond-the replicated unit would be valued on an amortized cost
basis because it's a bond. The cash instrument, the treasuries, are also valued on an
amortized cost basis for statutory reporting. And the derivative component would
also be valued on an amortized cost basis in this case.
14 RECORD, Volume 24
ln the other example-replicating the performance of an equity index-such as the
S&P 500, replicating would be on a market value basis. The cash instrument, in this
case, treasuries, would be valued on an amortized cost basis. And the derivative
would be valued on a market value basis. This gives you a feeling for how the rules
are developing right now for valuing the derivative component of a replication
transaction.
The next major item of the emerging regulatory framework is reporting. lt's likely
that a version of these reporting rules will be adopted at the upcoming NAlC
meeting for incorporation into the 1999 annual statement. lf that takes place, it's
possible that the other elements will fall into place and derivatives for replication
purposes will be recognized some time in 1999 or 2000.
The reason that's so important is that states are adopting the NAlC model investment
law with a proviso that the derivatives used for replication purposes will be
permitted as soon as the NAlC adopts a comprehensive framework dealing with
accounting, reporting, and RBC issues.
This report will be incorporated in Schedule DB, Part F. The key is that each of the
three pieces are identified. Then we have a description of the replicated or
synthetic asset. lf we were trying to replicate a lower-rated corporate bond, here's
the description of the replicating unit. The NAlC designation assigned to the
replicating unit would be a class 2, an SVO2 Bond with a value of $1,000. Then
Schedule DB starts taking apart the transaction into its various pieces. The
derivative instrument component is a default risk swap. The cash instrument is a
U.S. Treasury note. Finally, in designating the underlying cash instrument, as a
government bond, it is exempt.
One of the changes that might take place will be to incorporate within the various
components, not only descriptions, but also fair values. We would have a fair value
of the Treasury note, the swap, and the synthetic or replicated unit. The reason
we're moving in that direction goes back to the fact that the effectiveness
measurement is a key element to this whole project. As regulators, we want to
know that what you say you're replicating is what you are replicating and that the
risk characteristics are what they are claimed to be. The initial effectiveness
measurement tool was that, at time of inception, the fair value of the two
components-the cash market component and a derivative component-were equal
to the fair value of the replicated or synthetic asset. That idea has been pushed
along to require that as an effectiveness test throughout the life of the replication
transaction. To enable monitoring, we're moving towards incorporating fair values
within the two components and the replicating unit also.
15 Weird Science: Dissection of Derivatives
Another element of the complete package is asset valuation reserve (AVR) and RBC.
We've done some work on the AVR component (see Table 1), but the RBC format
has not been discussed yet.
TABLE 1DRAFT—MAY 5, 1998
ASSET VALUATION RESERVE (CONTINUED)BASIC CONTRIBUTIONS, RESERVE OBJECTIVE AND MAXIMUM RESERVE CALCULATIONS
REPLICATIONS (SYNTHETIC ASSETS)(1)
Position Number
(2)
Type (R/C)
(3)
CUSIP
(4)
Description of Asset(s)
(5) NAIC Designation
or Other Description of
Asset
(6)
Value of Asset
(7)
AVR Basic
Contribution
(8)
AVR Reserve
Objective
(9)
AVR Maximum Reserve
1 R CW
Fixed Rate Bond ABC Bond
Class 2 Class 2
$1,000,000 2,000 6,000 10,000
2 R
R
CW
Common Stock Portfolio
Five Year Treasury Coupon Strip U.S. Treasury Bond
Unaffiliated Public C/S 1 (Exempt)
1 (Exempt)
927,000
80,000
1,000,000
0
0
0
185,400
0
0
185,400
0
0 3A R
CW ABC Corporate Bond U.S. Treasury Note
Class 2 1 (Exempt)
1,000,000 1,000,000
2,000 0
6,000 0
10,000 0
4 R CW
XYZ 8% Bond Portfolio ABC Bond Portfolio
Class 1 1 (Exempt)
1,000,000 1,000,000
500 -500
1,500 -1,500
3,000 -3,000
5 R CW
Variable Rate Note ABC Bond
Class 2 Class 2
1,000,000 1,000,000
2,000 -2.000
6,000 -6,000
10,000 -10,000
6 R CW
Dollar Denominated Bond Foreign Denominated Bond
Class 1 Class 1
1,000,000 1,000,000
500 -500
1,500 -1,500
3,000 -3,000
7 R CW
Emerging Market Bonds U.S. Corporate Bonds
Class 3 Class 1
1,000,000 1,000,000
10,500 -500
28,000 -1,500
40,000 -3,000
8 R CW
XYZ Convertible Bond XYZ Bond
Class 3 Class 3
1,080,000 1,000,000
11,664 -10,800
30,240 -28,000
43,200 -40,000
9 R CW
Convertible Bond (ii) Corporate Bond
Class 1 Class 1
1,100,000 1,000,000
550 -500
1,650 -1,500
3,300 3,000
10 R CW
Floating Rate Bond Convertible Bond
Class 1 Class 1
920,000 1,000,000
460 -500
1,380 -1,500
2,760 -3,000
11 R CW
Index Amortizing Security Floating Rate Non-Callable Bond
Class 1 Class 1
1,000,000 1,000,000
500 -500
1,500 -1,500
3,000 -3,000
12 R CW
Callable Bond Fixed Rate Bond
Class 1 Class 1
1,010,000 1,000,000
505 -500
1,515 -1,500
3,030 -3,000
13 R CW
Fixed Rate Bond Floating Rate Bond with an 8% Cap
Class 1 Class 1
1,010,000 1,000,000
505 -500
1,515 -1,500
3,030 -3,000
14 R
CW
ABC Commercial Mortgage Loan
U.S. Treasury Note
Other Commercial, In Good Standing
1 (Exempt)
1,000,000
1,000,000
6,300
0
14,000
0
22,500
0 Line 0199999 Subtotal Default Component – Other Than Mortgage Loans 12,284 34,800 56,320 Line 0299999 Subtotal Defulat Component – Mortgage Loans 6,300 14,000 22,500 Line 0399999 Subtotal Equity Component – Common Stock 0 185,400 185,400 Line 0499999 Subtotal Equity Component – Real Estate and Other Invested Areas 0 0 0 Total 19,184 234,200 264,220
General Account – 53a and Separate Account – 31a
ln dealing with the transaction involving the replication of a lower-rated corporate
bond, look at example 3A. This report will be part of the annual statement that
provides information on the calculation of the AVR. You will be reporting both
components of the transaction. "R" stands for replicated unit, and "CW" stands for
the cash component of the transaction, with "W" indicating that there is no residual
16 RECORD, Volume 24
asset risk retained by the insurer in this transaction. ln fact, the swab transaction
would swap out any default residual risk if there was any. Because we're dealing
with a Treasury note, you don't have that. But if this was a SVO1 being swapped
into an SVO5, that would be a consideration when you're dealing with the cash
component.
Then you would identify both the cash component and the replicated component,
and provide the other information needed to do the calculations in this worksheet.
The basic idea of this worksheet is to calculate the additional AVR and eventually
the additional RBC of the replicating unit relative to the cash market instrument.
The worksheet shows several examples. ln some cases, what's called the AVR
reserve objective for the replicating unit will be a positive number, whereas the
AVR reserve objective for the cash market will be reported as a negative number.
And the subtotals are the sum of the positive and negative numbers. Otherwise,
only the increased AVR or RBC would be coming out on the worksheet. And the
objective is to provide the additional AVR or RBC of a replicating unit relative to a
cash market instrument.
Another element to consider is the pure and simple financial analysis. How are
regulators going to view replicated transactions relative to their components?
Basically, we're going to be viewing the replicated or synthetic asset as a single
entity. Even though it's built up or manufactured by the insurer, we'll be viewing
that as the entity in a financial analysis.
What does that mean? ln the first place, because, in many cases, we're talking
about transforming an exempt or a high-rated security into a lower-rated security,
who's going to determine the credit quality of that lower-rated replicated
transaction? That's where the SVO comes into play. These transactions will be filed
with the SVO, which will be asked to assess the replicated unit and assign an NAlC
designation to it.
One of the stumbling blocks in trying to apply that principle is dealing with
transactions that involve a basket or index. With a replication of a single issue, the
SVO can use its normal analytical tools to evaluate it. But, if you're dealing with
baskets or indices in the replication transaction, it becomes more difficult and
complex, so right now the SVO is reluctant to take on that task.
The first draft of the final version of this whole project will only deal with single
securities, not replications, baskets, or indices. However, those are the types of
transactions the industry is most interested in. lt does allow for diversification in
any specific market. And the industry is working very hard to suggest rules that the
SVO could apply in a realistic and efficient manner.
17 Weird Science: Dissection of Derivatives
From an asset adequacy analysis standpoint, regulators will be looking at the
replicated unit as being the asset supporting a block of business. lf you're
replicating a lower-grade corporate bond with treasuries and a swap transaction, we
would expect to see the bond being modeled for asset adequacy analysis purposes,
as opposed to modeling the treasuries, and possibly disregarding the swap
transaction. This seems to be the case with many companies that do hold derivative
instruments. So we would be looking at the replicating unit, not the components,
for asset adequacy analysis purposes.
Another element of the complete package is that we would be viewing the
replicated unit as a unit and applying it against any of the applicable limits. So if
you are trying to replicate the S&P 500, you're replicating common stock. That
replicated unit would be counted against your equity limits in your state's respective
investment law.
Finally, looking at it from a unit standpoint, if you have any financial ratios that your
state of domicile happens to use, presumably the states will be looking at the
replicated unit in their financial analyses. lf they have a liquidity test, or any test
that's uses a high-grade government bond as an element of that test, the state
probably would not use the government bonds in the way they normally would.
They'd be using it as the replicating unit. lf you're replicating the S&P 500, they'd
be looking at the replicating unit or financial ratios. They would not be using the
underlying government bonds in any type of liquidity test or high-grade bond test.
Mr. Reddy: You mentioned that replication isn't officially permitted right now, but
it's coming. To what extent do you feel that it's already going on, perhaps being
disguised as hedging? Second, with regard to credit derivatives, l know there's a lot
of activity going on now. What is the current regulatory status of that? ls it
technically permitted to do a credit swap, for example, and how should it go on the
books?
Mr. Gorski: Let's deal with the credit derivative transaction first, because similar
questions come up all the time. lf you purchased a structured note that has a credit
derivative embedded in it, that's perfectly legal. There's no problem with that
because that structured note is being filed with the SVO, which does a credit
analysis taking into account the credit derivative component of that structure note.
The problem emerges with stand-alone credit derivatives. And there are two paths
to take here. lf you're holding or engaged in a credit derivative transaction for
hedging purposes-hedging some credit risk that you may already have by entering
into and offsetting the credit derivative transaction-that poses no problem, because
derivatives are permitted for hedging purposes. lt's when you're taking on
18 RECORD, Volume 24
additional credit risk through a credit derivative transaction on a standalone basis
that problems occur.
The legal status of that type of transaction depends on your state of domicile. For
example, in lllinois, we adopted the new version of the model investment law, so
the transaction would be clearly prohibited. lt could not qualify via a basket either,
because the model investment law clearly says that you can't recognize derivatives
through a basket provision. ln other states, the answer may be ambiguous. Because
of the age of the law and the way it was written, the status may not be recognized.
But we realize that companies may be entering into a credit derivative transaction
for purposes other then hedging, so we're moving ahead as rapidly as we can with
the reporting end of this project.
Your other question was about companies entering into replication transactions
before the gun has sounded, so to speak. There are replication transactions and
there are replication transactions. lf you are replicating a floating-rate bond by
holding a high-grade corporate or a Treasury and entering into a swap transaction,
and there's no additional credit element to the swap transaction other than the
credit of the counterparty, you're not doing anything to change the credit risk of
that transaction. l don't think anyone would have much of a problem with that.
We would probably even recognize that as a hedging transaction in some fashion.
lt's where you are transforming the credit or asset risk of an instrument from an
RBC standpoint that regulators start to get concerned. l have no idea how many
companies are doing that. Reports in the trade press and by trade associations
indicate that some companies are, but their state law might permit it.
Mr. Reddy: l have a follow-up question for Charlene. We talked earlier about what
a bond is, how you can embed various elements into a bond structure, and that
bonds are rated by the SVO. lt's the synthetic asset creation idea that Larry talked
about. When you put something in a bond, it's clear that it can and should be
treated in a certain way. When we talk about credit derivatives, however, the
question is, are those credit elements being embedded in bonds today or are
insurers themselves using the stand-alone elements discretely to manage that risk?
We know that depository institutions, to a substantial degree, have been using credit
derivatives on a stand-alone and discrete basis. ln my insurance industry support
work, some groups have polled the industry, asking brokers, dealers, and others
about this practice. At this time, we haven't found more then a handful of
transactions being done by insurers on a stand-alone basis. We know there's a lot
of interest in it, though. Do you have any sense that has changed?
Ms. Barnes: No, because credit risk is the stock and trade of insurance companies.
And they are very comfortable taking it through most of the fixed income
19 Weird Science: Dissection of Derivatives
instruments that they normally deal with. They're not looking to separate it from
the rest of the risks associated with the bonds. Also, companies have expressed
some interest in private placement risk, even on a nonstand-alone basis, on their
existing assets. Some insurance companies see a private placement as having more
value than the publicly traded debt and want protection from the credit exposure.
They look at it as an arbitrage opportunity, and we're seeing a lot more of that.
Things come across my desk saying, for example, "We're willing to sell six months
worth of AT&T for two basis points" or something similar. The company is taking a
very short-term position and wants to get rid of something immediately. l've never
seen any interest from any insurance company in that type of transaction.
Mr. Gorski: About a year-and-a-half ago, some publication had a report on
derivative strategy that seemed to imply about 25 life insurance companies were
active players in the derivatives market. The article didn't clarify whether the
companies were using stand-alone derivatives or structured notes, or whether they
were taking on risk or hedging it away. However, it raised questions and, after
much discussion, the response came back consistent with what Charlene said.
However, it's still a nagging question, because at least once a week l get a phone
call from someone, that, reading between the lines, give me the feeling that
companies are doing something different from what Charlene is advocating. So l
still have a regulatory skepticism.
Mr. Ste�en P. Miller: Your definition of a hedge implies that it is something that
reduces risk. The FAS133 has a much narrower definition of a hedge, which seems
to exclude most of what insurance companies call hedges, including caps and
swaps for their single premium deferred annuity portfolios. Does anyone have any
comment about how that will affect the insurance industry's use of derivatives for
fixed annuities and GlCs?
Ms. Barnes: lt's not good. There are so many accounting issues, l could go on
about them for quite a while. And some of them seem very complicated to me.
The FAS doesn't have the resources to get into the absolute nitty-gritty of every issue
specific to insurance companies, and this does pose a problem.
Mr. Gorski: That's an interesting take on that question. As we were developing the
model investment law and the rules for recognition of derivatives for hedges,
regulators were being told that our rules were more limiting than rules in other
arenas. And now l'm hearing that maybe it's the other way around.
Mr. Anderson: This underscores the different paths that are being taken by GAAP
and statutory accounting. lt is truly remarkable. The unitary version we described
20 RECORD, Volume 24
in terms of the statutory approach serves the purpose of going along with legal
investment limits and so on. But we can't underestimate the impact of GAAP
accounting on companies. And certainly that has to influence behavior.
Mr. Erin Dandridge Cole: l am curious about these replicating transactions. lt
seems as though Schedule DB is trying to move the derivative piece, if it's a stand-
alone, into bond treatment. ls that so?
Mr. Gorski: The replicated unit is going to be treated as a bond or stock or what
have you. The individual pieces will be reported as they normally would be. The
cash component of the transaction would be reported in Schedule D and the
derivative portion would be reported in Schedule DB as an option or a future. But,
there will be a new exhibit "F" in Schedule DB that brings the pieces together so the
regulator can understand how the derivative is being used in conjunction with the
cash market instrument.
The AVR or the RBC piece will look at the replicated unit relative to the cash market
instrument and come up with the additional RBC. lt's walking a fine line between
viewing the replicated unit as a single entity versus viewing its pieces. lt's a bit of
both. But, if a company is viewing it as a replication transaction, we think our
regulatory framework should look at it the same way.
Mr. Cole: ls this happening because they're trying to get the NAlC to approve
putting on such derivatives? lt seems that companies stand to lose at least as much
by having the replicated unit treated as a package deal. lf a highly liquid bond that
provides some piece of the value of the total transaction is paired with a U.S.
Treasury bond, they're going to move the whole transaction down, which has some
obvious negative implications. lnstead, you could carry on your regular investment
process by throwing on the derivatives, and perhaps hold a lower-quality bond for a
longer duration, as you suggested. Why do they need to lump them together?
Mr. Gorski: lf it was just a lower duration bond that would be for hedging
purposes, these rules wouldn't apply. They only apply when the combination
transaction changes the characteristics of a bond from a RBC standpoint. Let me
respond to your initial question about whether the industry's desire to get this
additional use of derivatives recognized by regulators is causing everyone to follow
this path. When the industry suggested derivatives for replication purposes six or
seven years ago, we immediately objected to the things that we already talked
about. And this is the path that was agreed upon to remove those objections.
21 Weird Science: Dissection of Derivatives
Mr. Cole: So you're saying that's why they're going through this big effort to be
able to use a credit swap, for instance, because the NAlC won't allow companies to
do that unless they say it's trying to replicate something else?
Mr. Gorski: Right.
Mr. Cole: lt seems unduly rigid.
Mr. Gorski: That's the nature of regulation.
Mr. Sean Patrick Casey: l'm starting to deal with some asset/liability management
issues in our overseas operations. ls there much of a market overseas for interest
rate derivatives? ln particular, l'm looking at �apan, Korea and Taiwan. Also, if
these countries aren't that advanced, are they very efficient at this point?
Ms. Barnes: Do you mean within the country or dollar to yen?
Mr. Casey: Within the country.
Ms. Barnes: lt depends on the country. There are certainly some well-developed
markets. Swap markets are very well developed in �apan. The smaller the
economy, the less developed it's going to be. But, generally, it's doable. The
important considerations are how big the economy is, how much public debt there
is, and how long the yield curves go out. For example, Thailand's yield curve only
goes out five years, so you're not going to buy into very liquid 10-year swap market.
But they're certainly there.
Mr. Reddy: Larry, l have a question consistent with a replication project and putting
into place appropriate RBC treatment for these synthetic assets. Wouldn't it also be
appropriate to give a credit in cases when hedging is done for RBC purposes? l see
it as replication in reverse because you're laying off risk. Wouldn't it make sense to
provide some kind of credit for RBC where hedging is done?
Mr. Gorski: That's an excellent question. About three years ago, we actually had
that project on the agenda of the lnvested Asset Working Group. At the same time,
the industry was pushing for the replication project. We asked the industry which
project do you want us to take on first, recognition of risk reductions through
hedging transactions or replication transactions? The industry wanted us to tackle
replication first, so we're giving it the first priority. We'll be going back to the other
project as soon as this is wrapped up.
22 RECORD, Volume 24
CHART 1LARGEST INSURANCE COMPANY USERS OF DERIVATIVES
CHART 2OFF BALANCE SHEET APPLICATIONS